CAPACIDADES TECNICO PRODUCTIVAS

Formica aquilonia exhibited significant repellent responses to floral volatiles of 3/7 plant species tested (Figure 2.7), with all three being native to Britain. This supports previous work identifying floral ant-repellence as a common phenomenon in temperate angiosperms (Willmeret al.2009). For two further species ants spent noticeably less time on average in the floral volatile treatment airspace; however, the sample size was lower for both of these and they were not significant.

This brings up an important point about the olfactometer assays and associated statistical testing. When determining the magnitude of repellent effects, the mean % time in treatment airspace can be somewhat misleading, particularly as often around one to three ants per floral assay spent the entire duration of recording at one or other side of the olfactometer. As mentioned in the Methods and Materials (Chapter 2), the randomisation test accounts for this because when calculating significance, multiple replicates producing a small but consistent repellent effect will have much more weighting than just a few replicates with times at the extremes of the range.

Two of the three species eliciting repellent responses -Sambucus nigra (elder) andDigitalis purpurea (foxglove) - were chosen for olfactometer testing after observing that they had interactions withLasiussp. (probablyL. niger) ants in the field. Lasiussp. tend aphids on both

S. nigra (pers. obs.) and D. purpurea (Ballantyne 2011). On D. purpurea they additionally forage for floral nectar. S. nigra has extrafloral nectaries (EFNs) (Fahn 1987); however, the flowers contain no nectar (Atkinson & Atkinson 2002) and I did not observe any visits by ants to flowers. The incidence of floral repellence as assayed here withF. aquiloniacorrelates with the pattern of ant visitation in the field. F. aquilonia spent significantly less time in the treatment airspace for floral volatiles ofS. nigrabut had negligible responses toD. purpurea

(Figure 2.7). Information from field observations were however fromLasiussp. rather thanF. aquilonia, but as Lasius niger has previously been found to exhibit repellence related behaviours to floral volatiles from some plants also eliciting these behaviours inF. aquilonia

(Willmeret al.2009), it is possible the floral repellence detected withF. aquiloniacould help explain the distribution ofLasiussp. on these two plants.

Although ants are not as widely diverse in Britain as in other ecosystems, British plant species still have frequent interactions with ants, in particular through ant-attraction via EFNs and/or to ants tending aphids (e.g. Mahdi & Whittaker 1993; Oliveret al.2007). Plants occupied by aphid-tending or EFNs feeding ants may benefit from this interaction through protection against herbivores (Styrsky & Eubanks 2007), but any encounters with flowers are potentially detrimental to the plant (Galen 1999; Galen & Butchart 2003). Therefore, as in other ecosystems, plants stand to benefit from possessing ant-repellent floral volatiles.

In this example, as well as more generally, repellent floral volatiles are a particularly useful mode of preventing the potentially damaging effects of ants on flowers. Unlike morphological deterrents, repellent floral volatiles are able to act specifically where and when they are needed (Willmeret al.2009), are not subject to simple circumvention (e.g. Junkeret al.2011), and are less costly than morphological deterrents. This efficiency and specificity of repellent floral volatiles allows plants to moderate their interactions with ants, maintaining beneficial aspects of the interaction and also tipping the balance more towards favouring the establishment of new beneficial interactions that would otherwise not be possible.

It is interesting that plants with frequent interactions with ants (such asD.purpurea) do not necessarily have floral ant-repellents (Figure 2.7). Ballantyne (2011) found no evidence that bumblebee pollinators were deterred byD. purpureaflowers previously visited byL. niger, and it may be that costs of ant visits toD. purpureaflowers are minimal, thus correlating with the observed lack of repellence.

Whereas volatiles fromD. purpurea flowers elicited no repellent response fromF. aquilonia, the ants significantly avoided floral volatiles ofLinaria purpureain the same family, indicating (as in previous investigations) that there appears to be no clear phylogenetic pattern to floral ant-repellence.

The susceptibility of different floral forms to ant visits may be a predictor for floral ant- repellence, as there is evidence that various ant-deterring traits may trade off against each other (Willmeret al. 2009; Junkeret al. 2011). Centranthus ruberhas narrow corollas that would restrict ant access and correspondingly had no detectable repellent effect. On the other hand,Armeria maritima plants grow very close to the ground and have open inflorescences leaving them more susceptible to ant exploitation. The significant ant-repellent effect identified in A. maritima may be a consequence of this vulnerability. Linaria purpureawas unusual in this respect in that it elicited ant-repellent effects and also has a closed floral morphology.

Ants spent less time in the treatment airspace for both Hesperis matronalis and Buddleja davidii; however these data were not significant. ForH. matronalisalthough the mean time in treatment airspace was lower than 50%, a combination of small sample size and equipment

bias when testing meant that reliable conclusions cannot be drawn from this finding. The assay with B. davidii had a particularly low sample number so although there were still indications of a repellent effect, this was not significant (Figure 2.7). More replicates would be useful here asB. davidiiis very invasive in the UK and it would be interesting to investigate if possession of ant-repellent floral volatiles have a role to play in this.

Although floral volatiles from several of the plant species tested in the olfactometer were found to be repellent toF. aquilonia, this may not necessarily translate to repellence in the wild. F. aquiloniadon’t show a large polymorphism within the worker caste; however, there are differences in sizes between the workers (Sorvari & Hakkarainen 2009) and these may reflect polyethism (different behavioural roles) in workers (Wright et al. 2000). Apparent repellent floral volatiles may cause a different reaction in a worker specialised for foraging to one with the role of soldier, but no attempt was made to record these differences in the workers tested. Thus, the potential inclusion, in olfactometer trials, of workers unlikely to come into direct contact with flowers in the wild may obscure the reaction present in the wild. The reaction of ants to particular volatiles in the olfactometer is also likely to depend on their behavioural state. Previous work has suggested that elicitation of aggressive behaviours on exposure to floral volatiles can correlate with repellence (Willmer et al. 2009) and so it is possible that repellence from volatiles in the olfactometer is eliciting behaviours from the aggression-panic spectrum of alarm behaviours (Hölldobler & Wilson 1990) as suggested by the formic acid assays above. If this is the case with the floral volatiles, then a panicked, repellent response in the olfactometer may not translate to a similar response if the ant is in its own territory, where it may react with a more aggressive behaviour. Testing in the olfactometer was also carried out without any food source and the presence of this is likely to change the response of an ant to repellent volatiles. If repellent floral volatiles were present in a flower also possessing high nectar volumes, it is possible that ants may be able to overcome the repellence in order to reach the nectar.

Finally, while testing ants individually in the olfactometer helps avoid magnification of any potential effect through one following pheromone trails of another (Willmeret al. 2009), in the field ants often forage in association with many others (Dornhaus & Powell 2010). The reaction of an ant to a potentially repellent volatile encountered along an already established

pheromone-based trail is likely to be different to that exhibited by a solitary ant foraging in a new location. Further investigation into how repellence varies between these conditions would also be useful for further understanding this effect.